New Lists and Seasoned Firms: Fundamentals and Survival Rates. Eugene F. Fama and Kenneth R. French * Abstract

Transcription

1 First Draft: March 2001 Revised: July 2002 Not for quotation Comments solicited New Lists and Seasoned Firms: Fundamentals and Survival Rates Eugene F. Fama and Kenneth R. French * Abstract The class of firms eligible for public equity financing expands dramatically in the 1980s and 1990s. After 1979, the rate at which new firms are listed on major U.S. stock exchanges jumps from about 140 to near 600 per year. The characteristics of new lists also change. Cross-sections of profitability and growth become progressively more disperse, profitability becomes more left skewed, and growth becomes more right skewed. The result is a sharp decline in new list survival rates. The flood of new lists eventually causes similar but more subdued changes in the characteristics of seasoned firms, with a corresponding decline in survival rates. * Graduate School of Business, University of Chicago (Fama), and Tuck School of Business, Dartmouth College (French). We gratefully acknowledge the helpful comments of Frank Easterbrook, Owen Lamont, Kenneth Lehn, Jonathan Macey, Richard Roll, Hans Stoll, and seminar participants at UCLA. We thank Jay Ritter for giving us his list of initial public offerings.

2 The market for publicly traded equity is the heart of a modern capitalist system, signaling the terms on which investors are willing to bear residual corporate risks. The market for newly listed firms is in turn a bellwether for the public equity market. It is the point of entry that gives firms expanded access to equity capital, allowing them to emerge and grow. Examining the characteristics of newly listed firms can provide interesting information about changes through time in the kinds of firms that are viable candidates for public equity financing. The issue is important. In a perfect capital market (that is, absent monitoring costs and other frictions), investment is efficient: all wealth-creating projects are publicly financed, and their risks are efficiently shared among investors. But when frictions cause some profitable projects to be financed privately, or not undertaken at all, investment and risk sharing are inefficient relative to the zero-frictions optimum. If security prices are rational, evidence that the class of firms that are publicly financed broadens through time is evidence that demand conditions (the tastes of investors for different types of risk) or supply conditions (monitoring costs or other frictions) have changed. To the extent that the changes are due to supply conditions, the efficiency of investment and risk sharing improve. Fama and French (2001) document that the rate of new listings, largely on NASDAQ, explodes after 1979, from about 140 to near 600 per year. After 1979, on average about ten percent of listed firms are new each year. Our earlier paper examines the characteristics (profitability and growth) of new lists only in the listing year and not in much detail. Here we develop a detailed picture of the profitability and growth of the NYSE-AMEX-NASDAQ new lists of for the first five years after listing. And we examine how changes in the characteristics of new lists during the sample period affect whether they survive, disappear in mergers, or are delisted for poor performance. Our results about the evolving characteristics of firms that are viable candidates for public equity financing are easily summarized. The key words are dispersion and skewness. During , when new lists are abundant, the cross-section of new list profitability drifts down, and the drift is stronger in the left tail, that is, toward lower profitability. In contrast, the percentiles of new list growth drift up, and the drift is stronger in the right tail, toward more rapid growth. New lists eventually become seasoned

3 firms, and the profitability and growth of seasoned firms show subdued versions of the patterns observed for new lists; profitability and growth become progressively more disperse, profitability becomes more left skewed and growth becomes more right skewed. And we emphasize that although the process accelerates after 1994 (when high-tech and internet-related new lists are abundant), the increasing incidence of low-profitability high-growth firms is a long-term phenomenon, evolving slowly over the last 20 years. The drift in profitability and growth has a big effect on survival rates. The probability that a seasoned firm continues to trade beyond the next ten years falls from 59.5% for the cohort of 1973 to 44.3% for the 1991 cohort. The probability that a new list survives its first ten years falls further, from 64.4% for the 1973 cohort to 37.0% for the 1991 cohort. Rates of disappearance in mergers do not trend much during the sample period; on average, about one-third of the seasoned firms and one-fourth of the new lists of a given year are absorbed within ten years in mergers. The decline in survival rates is thus due to delistings for poor performance. The ten-year delist rate rises from 13.5% for the seasoned firms of 1973 to 21.7% for the 1991 cohort. The ten-year delist rate for new lists rises much further, from 14.4% for the 1973 cohort to 40.2% for the cohorts of Thus, about two in five of the new lists of are delisted within ten years for poor performance. For both new lists and seasoned firms, most of the changes in the cross-sections of profitability and growth during trace to small new lists. Since large firms account for the lion s share of most economic aggregates, one might argue that changes in the characteristics of small new lists are unimportant. The changes are, however, important for understanding the market for listed firms, that is, the kinds of firms that are viable candidates for public equity financing and thus unrestricted risk sharing. In essence, our results say that that changes in demand or supply conditions lead to increased sharing of the risks of firms with low profitability and high growth, a combination that produces a large dose of unhappy outcomes. Again, to the extent that this broadening of the kinds of firms publicly traded is due to supply conditions (reductions in the costs of trading and holding such firms), the efficiency of the economy s investment and risk sharing are enhanced as long as risks are properly priced. 2

4 Finally, our results are related to those of Campbell, Lettau, Malkiel, and Xu (2001). They find that the idiosyncratic dispersion of individual stock returns increases through time and the increase is largely due to small firms. Our results suggest that the source of the higher return dispersion is increased dispersion of profitability and growth, which largely traces to the post-1979 flood of small new lists. We begin (section I) by detailing the rate of new listings during Section II examines the average profitability and growth of new lists and seasoned firms. The central evidence on the evolution of the cross-sections of profitability and growth is in section III. Section IV examines survival rates, and section V studies the links between survival rates and the profitability and growth of new lists and seasoned firms. Section VI concludes and offers some perspective on whether the expansion of the class of publicly traded firms during is due to changes in demand or supply conditions. I. New Lists: Counts and Size Figure 1 shows annual counts of combined NYSE, AMEX, and NASDAQ new lists for To be in the sample, a firm must be on the files of the Center for Research in Security Prices (CRSP) and have a share code of 10 or 11 (ordinary common shares), so ADRs and closed end mutual funds are excluded. We define a new list as the first appearance of a firm on CRSP. Thus, our new lists do not include firms that switch from one of the three exchanges to another. We also exclude tracking stocks, spin-offs, and firms that go public after going private. The tests start in 1973, the beginning of the CRSP NASDAQ period. Prior to 1973, newly public firms typically trade over the counter (OTC, not covered by CRSP), and new listings on the NYSE and AMEX (covered by CRSP) are mostly seasoned firms. NASDAQ absorbs most of the OTC market, and for the post-1972 period, the CRSP files provide a rather complete picture of publicly traded firms. We examine two types of new lists: initial public offerings (IPOs) and non-ipos. A firm is defined as an IPO if it is in the IPO sample provided by Jay Ritter (an updated version of that in Loughran and Ritter (1995)) and its CRSP listing month is no more than ten months after its IPO. Non-IPO new lists include (i) a small set of firms that are in the Ritter IPO sample but are listed on one of the three 3

5 exchanges more than ten months after their IPO, and (ii) a larger set of new lists not in the Ritter IPO sample. The latter are primarily (i) tiny firms that trade OTC (pink sheets) and do not make it to one of the three exchanges soon after their IPO, and (ii) some financial firms that are missing from the Ritter IPO sample but would qualify as IPOs under the ten-month rule. Since our main focus is the economic fundamentals of non-financial firms, we do not attempt to more accurately allocate financial new lists between IPOs and non-ipos. One can argue that, to study the kinds of firms that are viable candidates for public equity financing, the ideal sample is all IPOs. Our IPOs include only those listed quickly on one of the three exchanges (primarily NASDAQ). We miss IPOs that initially trade OTC. Many of these eventually appear in the sample of non-ipo new lists, and this is why we use IPOs and non-ipo new lists to make inferences about the characteristics of new publicly traded firms. There is, however, a survivor bias in this approach; (financials aside) non-ipo new lists include only those initially unlisted IPOs that are relatively successful and so eventually make it to one of the three exchanges. It is thus likely that our inferences about the kinds of firms that qualify for public equity financing are conservative. There is a more aggressive justification for our approach. One can argue that during , IPOs not traded on the NYSE, AMEX, or NASDAQ are typically illiquid and so do not get the benefits of unrestricted risk-sharing. In this view, during , an exchange listing is the better signal that a firm qualifies for unrestricted risk sharing. Thus, examining new lists (IPOs and non-ipo) is a sound approach. Moreover, changes in the characteristics of new lists related to changes in listing requirements are not a problem if the exchanges compete for listings. Competition implies that listing requirements are themselves the result of demand conditions (the tastes of investors for different types of risks) and supply conditions (monitoring costs, information costs, and other frictions) that determine the types of firms that qualify for unrestricted risk sharing. Two facts are apparent in Figure 1. First, after moderate increases from 1973 to 1979, new lists surge from 220 in 1979 to 438 in 1980 and 620 in After 1979, only four years have less than 400 new lists. For , new lists average 598 per year (Table 1). There are not many IPOs in the 4

6 period (on average 30 per year); most new lists are non-ipos (119 per year). But during , IPOs are more numerous than non-ipo new lists, averaging 372 per year, versus 226 for non-ipo new lists. We can also report that more than 90% of the new lists of are on NASDAQ. Table 1 summarizes the size of annual new lists. We measure size as market capitalization, ME, stock price times shares outstanding. The table shows the averages of the yearly average NYSE ME percentile of new lists and the average local (listing) exchange ME percentile. When compared to firms on their respective exchanges (NYSE, AMEX, or NASDAQ), new lists are on average medium sized. For the average local exchange ME percentile of all new lists, 50.3, is just a bit above the local exchange median. There is only one year, 1976, when the average local exchange ME percentile of all new lists is below 40. IPOs are on average larger than non-ipo new lists. For the average local exchange ME percentile of IPOs is 57.9, versus 43.1 for non-ipo new lists. But in absolute terms new lists are typically tiny. The average NYSE ME percentile of the new lists of is 12.1 (Table 1). IPOs are on average at the 15.1 NYSE ME percentile, versus 8.5 for non-ipo new lists. The surge in new lists after 1979 is not associated with a decline in size. The average NYSE and local exchange ME percentiles of new lists (IPOs and non-ipos) increase from to (Table 1). More important, during there is a progressive thinning of the extreme left tails of the size distributions of new lists (IPOs and non-ipo). For example, the proportion of new lists below the 20 th NYSE ME percentile falls from 85.3% for to 70.1% for (Table 1) and the proportion below the 10 th percentile falls from 75.2% to 50.0%. Thus, the evolution of new list fundamentals documented below (increasing left skewness of profitability and right skewness of growth) is not due to higher frequencies of the tiniest firms. This point is important when we later (in the concluding section) discuss whether the broadening in the types of firms publicly traded during is due to demand or supply conditions. Finally, with the high rate of new listings for , on average around ten percent of listed firms are new each year (Table 1). New lists are mostly small, however, and despite the explosion in their 5

7 numbers, the fraction of the aggregate market value of listed firms accounted for by annual new lists is also small, averaging 2.08% for and never exceeding 4.1%. II. Average Profitability and Growth Table 2 summarizes the evolution of fundamentals (average profitability and growth rates) for the non-financial new lists of for the first five years firms are listed. The fundamentals are also shown for matched cohorts of seasoned non-financial firms. Seasoned firms are defined as NYSE, AMEX, and NASDAQ firms listed more than five years, and firms already on NASDAQ at the end of the CRSP startup period (April 1973). Fundamentals for all firms (not shown) are similar to those of seasoned firms. The samples in Table 2 are smaller than in Table 1 because financial firms are excluded from Table 2 and because the fundamentals are sometimes missing from the Compustat data source. The fundamental variables in Table 2 are ratios of aggregates. For example, the profitability in year t+τ of the new lists of year t, E t+τ /A t+τ, is the ratio of aggregate earnings before interest for t+τ of the new lists of year t divided by their aggregate t+τ assets. In effect, then, we measure fundamentals as if the new lists of year t are a single firm. Equivalently, ratios of aggregates are size-weighted averages of the ratios for individual firms. For example, the estimate of new list profitability weights the profitability of an individual year t new list by the ratio of its t+τ assets to the total t+τ assets of all year t new lists. Size-weighted averages give more weight to larger firms and so might not provide a picture of fundamentals for the typical firm. But Table 2 is just an introduction to salient characteristics of fundamentals for new lists and seasoned firms. We later examine time-series of cross-sections of profitability and growth, which are the core of our story. Finally, our estimates of fundamentals for t+τ can cover only the firms of year t with Compustat data for t+τ. Firms that disappear before t+τ are not covered in the estimates. 6

8 A. Growth New lists grow faster than seasoned firms. The growth rate of total assets, da/a = (A t -A t-1 )/A t-1, for the seasoned firms of averages 10.1% per year. The average first-year growth of IPOs is 78.7%, declining rapidly to 20.0% in the fifth listed year. The average first-year growth of non-ipo new lists is 19.9%, with no systematic tendency in subsequent years. Thus, IPOs initially grow faster but eventually converge on the growth rates of non-ipo new lists, which in turn grow about twice as fast as seasoned firms. These results suggest that non-ipo new lists come into the sample while still in a high growth phase but after the period of extreme initial growth typical of IPOs. B. Profitability In the listing year, IPOs are on average more profitable than non-ipo new lists. This is true for the full sample period and all subperiods in Table 2. For the full sample period, IPO profitability declines in the years after listing, but E/A rises for non-ipo new lists. As a result, the two groups have similar average profitability after the second listed year. Thus, as they age, the profitability and growth of IPOs come to look more like those of non-ipo new lists. More interesting, for the full period, IPOs are on average more profitable in the listing year than seasoned firms. But after the listing year, IPO profitability falls and they become progressively less profitable than seasoned firms. Figures 2a to 2c give year-by-year details on the average profitability of new lists and seasoned firms in the first, third, and fifth listed years. The year on the horizontal axis is always the listing year. For example, plotted at 1973 in Figures 2a to 2c are the 1973, 1975, and 1977 average profitability of the new lists of 1973 and the seasoned firms of The figures thus compare the evolution of average profitability for cohorts of new lists and seasoned firms. The average profitability of new lists in Figures 2a to 2c varies much more across cohorts than the average profitability of seasoned firms. Despite this high volatility, there are clear patterns in the relation between new list profitability and that of seasoned firms. The first-year average profitability of IPOs (in Figure 2a) is higher than the profitability of seasoned firms in all but two of the first 22 years, 7

9 from 1973 to In contrast, the first-year profitability of non-ipo new lists tends to be below that of both IPOs and seasoned firms throughout the sample period. After 1994 the first-year profitability of new lists (IPOs and non-ipos) falls progressively further below that of seasoned firms. The first-year profitability of IPOs goes negative in 1999, for the first time in the sample period. The first-year profitability of non-ipo new lists is negative in 2000, but the highly volatile first-year profitability of these firms is also negative in three earlier years. The decline in IPO profitability in the years after listing is evident in Figures 2b and 2c. In contrast to the higher listing year average profitability of IPOs in the years up to 1994, IPO average profitability in the third and fifth listed years tends to be below that of seasoned firms for cohorts after 1980 or perhaps And after the listing year, IPOs are not systematically more or less profitable than non-ipo new lists. In short, for at least the last 20 years of the sample period, new lists (IPOs and non- IPOs) are on average less profitable after their second listed year than seasoned firms. C. The Small-Firm Depression Fama and French (1995) document a sustained decline in the profitability of small firms relative to big firms in the 1980s. Figure 3 confirms that small firms (total assets below the NYSE median) are more profitable than big firms until Thereafter, small firms are less profitable than big firms. After narrowing in , the gap between big and small firm profitability widens dramatically. In 2000, average E/A is 6.4% for big firms and only 0.7% for small firms. Since new lists are mostly small, it is interesting to examine whether the relatively low profitability of small firms during the 1980s and 1990s is related to the flood of small new lists and the low profitability of new lists as they age. 1 Figure 3 confirms that much of the small-firm depression of the 1990s is due to new lists. Until 1990, the profitability of small new lists (defined as firms listed within the last five years with total assets below the NYSE median) is similar to the profitability of small seasoned firms. Thereafter, small 1 When we examine fundamentals (profitability and growth rates), we define size in terms of assets, not market equity. Defining size in terms of market equity tends to allocate firms that are large in terms of assets but have low profitability to the small group. As a result, small market equity firms tend to look more like weak firms than when size is defined in terms of assets. 8

10 seasoned firms are only a bit less profitable than big firms, but small new lists are much less profitable. In 2000, for example, average E/A is 6.4% for big firms and 5.5% for small seasoned firms, but it is a miserable -8.8% for small new lists. (And keep in mind that we can measure profitability only for firms that survive.) Skipping the details, the gap between the profitability of big firms and seasoned small firms narrows further if we restrict seasoned firms to those listed for at least ten years. In fact, with this tighter definition, seasoned small firms are actually more profitable than big firms in 1992, 1993, and It is clear that, with their sustained low profitability, the flood of small new lists in the 1980s and 1990s plays a substantial role in the sustained low profitability of all small firms. Finally, Jain and Kini (1994) examine the profitability of the IPOs of They find that when firms go public, median IPO profitability is higher than the median profitability of firms in the same industry. After the IPO, median profitability falls toward that of the industry-matched sample, and the median investment of IPO firms is higher than for the industry-matched sample. Mikkelson, Partch, and Shah (1997) report similar results for the IPOs of 1980 to This earlier IPO evidence is roughly similar to our results for the IPOs of the late 1970s and early 1980s, but it does not describe IPO performance later in our sample. In the 1990s, post-listing IPO profitability deteriorates to levels far below that of seasoned firms. And after 1994, even the first-year profitability of IPOs is lower than the profitability of seasoned firms. We also show that the new lists of that are not recent IPOs look much like aging IPOs; that is, they are less profitable but grow faster than seasoned firms. III. Cross-Sections of Profitability and Growth The average profitability and growth of new lists are not necessarily surprising, at least prior to the profitability plunge after Thus, it is not surprising that firms have initial public offerings when profitability is high and they have strong demand for equity capital to finance rapid growth. And if initial profitability is unusually high, it is not surprising that it falls in the years after listing, as growth causes firms to deplete their most profitable investment options. The interesting surprises, and the core of our 9

11 story, are in the changes through time in the dispersion and skewness of profitability and growth for new lists and seasoned firms. A. Profitability Figure 4a shows time series of the cross-sections (10 th, 25 th, 50 th, 75 th, and 90 th percentiles) of first-year profitability, E/A, for new lists. Until 1978, the 10 th to 90 th percentiles of new list profitability cover a rather narrow range (relative to subsequent years) and unprofitable firms are rare. Thereafter, all percentiles of new list profitability fall. For example, median first-year E/A for the new lists of 1978 is 0.115; it declines to in 1990 and in 1998, and then drops below 0.0 in 1999 and Thus, more than half of the new lists of 1999 and 2000 are unprofitable in their listing year. But the dominant trait of Figure 4a is the increasing dispersion in first-year profitability, due to increasing left skewness. The 25 th percentile of E/A falls from in 1978 to in 1990 and in The decline in the 10 th percentile is even more extreme, from in 1978 to in 1990 and in In short, during , when new lists are consistently abundant, firms with low indeed severely negative current profitability become progressively more acceptable candidates for public equity financing. Figure 4b shows annual cross-sections of profitability for firms listed in the previous five years. Cumulating over five years provides longer-term perspective on the performance of new lists. Though smoother, the percentiles of E/A for new lists of the previous five years are similar to those of first-year E/A. Thus, the increasing dispersion and left skewness of new list profitability are not just a listing-year phenomenon. Skipping the details, we can also report that increasing dispersion and left skewness are common to the evolution of profitability for both IPO and non-ipo new lists. Typically, more than 95% of new lists are small (assets below the NYSE median), so small firms dominate the distributions of profitability in Figures 4a and 4b. (Indeed, since they are so similar to Figures 4a and 4b, we do not include plots for small new lists.) Figure 4c shows that big new lists do not share the extreme profitability traits of their small counterparts. Though the percentiles of E/A for big 10

12 new lists of the previous five years decline a bit through time, the cross-sections are rather compact and do not show the strong left skewness that characterizes the distributions for all new lists. Averaging over , about ten percent of listed firms are new each year. How does this flood of new lists affect the cross-sections of profitability for seasoned firms and all listed firms? After they have been listed for five years, we reclassify new lists as seasoned. The cross-sections of profitability for all seasoned firms in Figure 5a become progressively left skewed during , but less so than for new lists. This is not surprising. Economic logic along with the evidence below on the low survival rates of new lists says that firms cannot sustain large losses indefinitely. As in the case of new lists, the profitability cross-sections for all seasoned firms are dominated by small firms. Figure 5b shows that the dispersion of profitability for big seasoned firms increases only a bit through time, and the distribution remains relatively compact and roughly symmetric. The median profitability of seasoned big firms is stable, in 1973 and in 2000, but the most profitable seasoned big firms become a bit more profitable, and the least profitable become a bit less profitable. The cross-sections of profitability for seasoned firms listed at least ten years (Figure 5c) show that much of the increasing left skewness of profitability observed in Figure 5a (seasoned firms listed at least five years) is driven by unprofitable new lists. Tightening the definition of seasoned firms reducing the impact of aging new lists has little effect on the right tail of the distribution; there is little difference between the 75 th and 90 th percentiles in Figures 5a and 5c. Tightening the definition does, however, dampen the distribution s left skewness. For example, the 10 th percentile of E/A for firms listed more than five years (Figure 5a) falls from 3.5% in 1978 to -11.4% in 1990 and -17.9% in The 10 th percentile for firms listed more than ten years (Figure 5c) falls only about half as much, from 3.6% in 1978 to -5.2% in 1990 and -9.1% in Figure 6 shows how new lists and seasoned firms combine to produce the cross-sections of profitability for all listed firms. From 1980 to 2000, the distribution of profitability for all listed firms drifts down and becomes increasingly skewed left, more so than for seasoned firms but less strongly than for new lists. This is not surprising, given the evidence on the evolution of profitability for new lists and 11

13 seasoned firms. Our point is that the mostly small new lists of play an important role in the evolution of the cross-section of profitability for all listed firms (the population of firms that get the benefits of unrestricted risk sharing) both because new lists are so numerous and because aging new lists are influential in the evolution of profitability for seasoned firms. B. Growth Rates There are also substantial changes in the growth characteristics of listed firms during the period of abundant new lists. Like profitability, the distribution of growth for all listed firms (Figure 7a) becomes more disperse. While profitability becomes more left skewed, growth becomes more right skewed. Median growth does not change much during , fluctuating around 10% per year (close to the size-weighted average growth of seasoned firms in Table 2). But firms with shrinking assets become more common. In 1973 about 10% of listed firms decrease in size from one year to the next; after 1981, typically more than 25% shrink from one year to the next. Increasing right skewness is, however, the obvious feature of the cross-section of growth rates. The 75 th percentile of da/a rises from 24.3% in 1973 to 42.3% in 2000; the 90 th percentile rises from 43.6% to 184.2%. Figure 7b shows that the increasing right skewness of asset growth for all listed firms is much subdued in the cross-section of seasoned firms (listed more than five years). Thus, the more extreme skewness observed for all listed firms is due to new lists. Indeed, to accommodate the increasing right skewness of da/a for new lists of the previous five years, the upper end of the scale in Figure 7c must be stretched to 6.0 (600 percent per year), versus 2.2 for all listed firms in Figure 7a. Figures 8a and 8b show that the cross-sections of da/a for IPO and non-ipo new lists are skewed to the right, but the asymmetry is more extreme for IPOs. And the right skewness of da/a observed for new lists of the previous five years (in Figures 8a and 8b) is dwarfed by the skewness of growth rates in the first listed year (in Figures 9a and 9b), especially for IPOs. One quarter of the IPOs in 1999 have one-year asset growth rates that exceed 1100 percent, and ten percent of the growth rates 12

14 exceed 3000 percent. The 75 th and 90 th percentiles of one-year asset growth rates for IPOs in 2000, 682 percent and 1375 percent, are smaller, but still extraordinary. Perhaps the extreme initial growth of IPOs is not surprising. Many firms go public because they are in a high growth phase and have strong demands for external equity financing. Toward the end of the sample period, many firms (especially internet related firms) go public early in their life cycles, when they have few tangible assets. Enormous asset growth in percentage terms is not surprising from a base close to zero. But Figures 8a and 8b show that strong right skewness characterizes cross-sections of new list (IPO and non-ipo) growth rates for the first five listed years. And the increasing right skewness occurs while there are fewer and fewer new lists in the extreme left tail of the size distribution (Table 1). Moreover, Figure 8c shows that many big new lists, with assets above the NYSE median, experience extreme growth toward the end of our sample. Each year from 1995 to 2000, one quarter of the big new lists of the last five years have annual asset growth rates above 65% and ten percent have growth rates above 145%. In 2000, the 10 th percentile for big firms listed in the last five years is 814%. Though right skewness is the dominant characteristic of new list growth, there is also interesting action in the left tail of the distribution. The 10 th percentile of first-year da/a for IPOs is always positive (Figure 9a); few IPOs shrink in their first listed year. And for IPOs, all percentiles of first-year da/a increase in the 1980s and 1990s; stronger first-year growth is a general characteristic of the IPOs of later years. In contrast, though difficult to see because of the extreme scale of the graphs, after 1981 about 25% of non-ipo new lists have shrinking assets even in the listing year (Figure 9b). Moreover, after 1981 about 25% of the IPOs of the previous five years have shrinking assets (Figure 8a). And if anything, the median and lower percentiles of asset growth for new lists (IPOs and non-ipos) of the previous five years decline after 1981 (Figures 8a and 8b). These results suggest that the increasingly low and left skewed post-listing profitability of the new lists of the last 20 years eventually produces a substantial fraction of new lists that begin to shrink soon after listing. In sum, during , more and more new firms with high growth and low profitability become viable candidates for unrestricted risk sharing via publicly held equity. Because these new firms 13

15 are plentiful, their characteristics eventually dominate the characteristics of listed firms (the population of firms that get the benefits of unrestricted risk sharing). But new lists are mostly small, and the changes they induce in the characteristics of listed firms are largely special to small firms. Profitability and growth also become more disperse for large firms, large-firm growth becomes somewhat right skewed, but large-firm profitability does not become noticeably left skewed. In general, the changes in growth and profitability for large firms are dwarfed by those for small firms, especially small new lists. And with the flood of new lists after 1979, more and more small seasoned firms are aging new lists with continuing high growth and low profitability. In the end, then, the central role in the changing characteristics of listed firms falls to new lists, especially small new lists. IV. Survival Rates Some of the changes in profitability discussed above may be a spurious result of accounting rules. For example, the high profitability of the early sample years may be due in part to the high inflation of the 1970s and early 1980s, which causes profitability to be overstated because earnings grow with inflation but assets are measured at historical cost. Another common story is that profitability is understated later in the sample period because firms invest more in intangible assets like R&D and human capital, which are expensed rather than depreciated over time. We doubt that vagaries of accounting can explain the major changes in profitability we observe the increasing left skewness of E/A for small firms and especially small new lists, which is not shared by big firms. In any case, there is a simple test. If the increasing skewness of profitability is a matter of accounting rules, it should not be associated with changes in survival rates. But if the left skewness of profitability is real, survival rates, especially for small firms and small new lists, are likely to decline through time. Table 3 summarizes average survival rates, specifically, percents of firms still trading after ten years and percents lost within ten years in mergers or through delisting for poor performance. Year-byyear details are in Figures 10 and 11. Survival rates indeed decline through time. For seasoned firms, the ten-year survival rate falls about 15 percentage points, from 59.5% for the 1973 cohort to 44.3% for the 14

16 cohort of 1991 (Figure 10a). Survival rates for new lists are close to those of seasoned firms early in the sample period, but new list survival rates fall more through time. The ten-year new list survival rate falls more than 25 percentage points, from 64.4% for the new lists of 1973 to 37.0% for the 1991 cohort (Figure 10a). Though there are no clear trends after 1981, survival rates fall more after 1973 for non-ipo new lists than for IPOs. The ten-year survival rate for IPOs falls from 56.7% for the cohort of 1973 (Figure 10b) to an average of 39.7% for the cohorts (Table 3). The decline for non-ipo new lists is from 68.7% to 30.6%. Thus, for (when new lists are plentiful and their profitability declines, becomes more disperse, and more left skewed), only about 40% of IPOs and 31% of non-ipo new lists survive more than ten years. Firms disappear from CRSP in mergers or because poor performance causes them to be delisted. Merger targets include strong and weak firms, so takeovers are ambiguous signals about performance. There is, however, no ambiguity about the poor health of firms delisted for cause. And, in line with the profitability evidence, new lists are more likely to be delisted for poor performance than seasoned firms. Figure 11a shows that the rate at which seasoned firms are delisted for poor performance increases through time; 13.5% of the seasoned firms of 1973 are delisted within ten years (Figure 11a), and the average ten-year drop rate rises to 18.4% for the seasoned cohorts of (Table 3). The drop rates for new lists, which are always above those for seasoned firms, rise sharply from 1973 to roughly Thereafter, the new list drop rates are too variable to identify a clear trend. The ten-year drop rate for the new lists of 1973, 14.4%, is a bit higher than for seasoned firms, but the average for the cohorts of is about twice as high, 40.2%. Thus, about one in five of the seasoned firms of is dropped within ten years for poor performance, versus two in five new lists. The 1973 IPOs and non-ipo new lists (Figure 11b) have similar ten-year drop rates, 13.3% and 15.0%, but the rate for non-ipo new lists rises more thereafter, averaging 47.5% for cohorts, versus 33.5% for IPOs 15

17 (Table 3). Thus, almost half of the non-ipo new lists of are dropped for poor performance within ten years of listing, versus (a still impressive) one-third of IPOs. 2 The profitability of big seasoned firms becomes only a bit more disperse during the 1980s and 1990s, and it does not show the left skewness observed for small seasoned firms (Figure 5b). Not surprisingly, then, the (always low) ten-year drop rate for big seasoned firms rises only slightly, from 1.7% for the cohorts of to 2.1% for (Table 3). The profitability of big new lists (Figure 4c) is more disperse than that of big seasoned firms but much less disperse and left skewed than that of all new lists (Figure 4a) and, by implication, small new lists. Thus, it is also not surprising that ten-year drop rates for big new lists, which average 7.1% for and 6.7% for , are higher than the rates for big seasoned firms but much lower than for small new lists. Mergers are the main exit route for big firms; 27.4% of the cohorts of big seasoned firms and 36.4% of big new lists are absorbed within ten years in mergers. In sum, the evidence on performance delistings conforms nicely with the profitability results. Specifically, with the surge in new lists after 1979, the cross-section of profitability for listed firms drifts down and becomes progressively more left skewed. These changes in profitability are mostly due to small firms and they are stronger for new lists than for seasoned firms. Likewise, performance delistings increase after 1979, the increase is primarily due to small firms, and it is larger for new lists than for seasoned firms. V. Profitability for Outcome Groups of New Lists The analysis above presumes that firms delisted for poor performance are from the fattening left tail of the cross-section of profitability. Table 4 provides evidence. The table shows the average profitability and growth of new lists and seasoned firms for one, three, and five years before each of the possible outcomes, survival, merger, or performance delisting. For example, current average profitability 2 Some readers have asked how new lists fare in the difficult markets of 2000 and The one-year delist rate for the cohort of 2000, 3.0%, and the two-year rate for the 1999 cohort, 9.2%, are not extraordinary. They are lower, for example, than five of the ten one-year rates and eight of the two-year rates for the cohorts of

18 is shown for portfolios of firms listed within the previous five years that delist for cause within the next one, three, and five years. Three merger portfolios are defined in the same way. And there are three complement portfolios that include new lists of the previous five years that survive at least through the next one, three, and five years. Table 4 also shows average profitability for seasoned firms (listed at least five years) that merge within, delist within, or survive beyond the next one to five years. The general decline in profitability from to hits different groups of firms at different times and to different extents. In the 1970s and 1980s, survival implies strong average profits for seasoned firms. IPO survivors are even more profitable and grow about three times faster than seasoned survivors. Seasoned survivors share the general decline in profitability from to , but the decline is stronger for IPO survivors. In the 1990s, when E/A averages 6.1% to 6.3% for seasoned survivors, it is 6.0% for IPOs that survive beyond the next five years, and only 4.6% for those that survive beyond one year. The decline in profitability shows up earlier among non-ipo new lists that survive. For , the average profitability of non-ipo new list survivors is similar to that of seasoned survivors. By the 1980s (and in 1990s), the average profitability of non-ipo new list survivors while still respectable is 1.5% to 2.0% lower than for seasoned survivors. Overall, however, survivors among new lists and seasoned firms tend to be profitable with strong growth, which is not surprising. Survivors are, however, declining proportions of all seasoned firms and new lists. The evidence on the profitability and growth of firms lost in mergers is more interesting and novel. Merger rates do not trend much during the sample period. During , on average about one in three seasoned firms and one in four new lists are absorbed in mergers within ten years (Table 3). But the characteristics of merged firms change. The seasoned firms of lost in mergers are about as profitable and grow nearly as fast as seasoned survivors (Table 4). But in the 1980s and 1990s, the average profitability and growth of merged seasoned firms fall relative to seasoned survivors. Thus, through time the merger arrow is aimed more at mediocre seasoned firms. (This is a likely explanation for the low pre-announcement stock returns of merged firms (Mitchell and Stafford (2000)).) 17

19 The picture for IPOs lost in mergers is a bit different. In the 1980s and 1990s, their average profitability and growth for years far in advance of merger are high, like those of IPOs that survive. Average profitability deteriorates as merger approaches, but growth remains strong. These results suggest that merged IPOs tend to be firms that continue to grow despite poor returns on investment. Non-IPO new lists lost in mergers also experience declining profitability in the years preceding merger. As expected, the biggest effects of the downward drift and increasing left skewness of profitability after 1979 show up in the expanding set of firms delisted for poor performance. They have terrible earnings for five years before delisting, and profitability deteriorates as delisting approaches (Table 4). New lists (IPOs and non-ipo) delisted for poor performance are less profitable than delisted seasoned firms. Consistent with the increasing left skewness of profitability, the poor profitability of delisted firms (seasoned and new lists) becomes more extreme from to And the delisted firms of have negative and declining profitability for five years before delisting. The average growth of firms delisted for poor performance is interesting. Delisted seasoned firms have low growth in the years preceding delisting and shrinking assets in the delisting year (Table 4). More surprising, but in line with the results for merged new lists, the new lists of delisted for poor performance on average grow strongly in the years preceding delisting, despite typically negative profitability; they do not begin to shrink until the delisting year (if then). Thus, the high delist rates of new lists tend to be the result of growth un-rewarded by earnings. These results suggest that new lists delisted for poor performance tend to be firms that (at least on an ex post basis) have wasted resources on unprofitable investments. And the suggestion is stronger than for merged new lists. VI. Conclusions After 1979, the rate at which new firms are listed on the major U.S. stock exchanges jumps from about 140 to near 600 per year. The profile of new lists also changes. Profitability and growth become progressively more disperse, profitability becomes more left skewed, and growth becomes more right skewed. The result is a sharp decline in new list survival rates due to delistings for poor performance. 18

20 The flood of new lists with low long-term profitability and high growth eventually causes seasoned firms to acquire subdued versions of the profitability and growth characteristics of new lists, with a corresponding decline in survival rates. The dramatic changes in the profitability and growth characteristics of listed firms during are largely due to small firms. The changes are nevertheless important for understanding the market for listed firms, in particular, the kind of firms that are viable candidates for public equity financing. Our results say that changes in demand or supply conditions lead to increased sharing of the risks of firms with high growth and low profitability, a combination that produces a large dose of unhappy eventual outcomes. And these changes occur slowly over the post-1979 period. They are not special to the hightech and internet-related new lists of last few years. The broadening of the kinds of firms publicly traded during may be due to changes in supply conditions, such as lower monitoring costs and reductions in other costs of holding and trading small, relatively unprofitable, but rapidly growing firms. If changes in supply conditions are responsible, and if new firms are properly priced, expanding the set of firms eligible for public trading enhances the efficiency of the economy s aggregate investment and risk sharing. There is, of course, controversy on the pricing issue. Behavioralists, like Ritter (1991) and Loughran and Ritter (1995), argue that IPOs are overpriced and yield abnormally low post-listing returns; in effect, there is too much investment in IPOs and their activities. Others, like Fama (1998) and Brav, Geczy, and Gompers (2000), argue that the way returns are risk-adjusted has a big effect on inferences about IPO pricing, rendering all inferences shaky. In a provocative recent paper, Schultz (2002) argues that because IPOs bunch in periods following high returns, the average return on the typical IPO is likely to appear low, even if IPOs are properly priced. Suffice it to say that the pricing issue remains open. Is the broadening of the types of publicly traded firms during due to demand or supply conditions? This is a big question, and we only offer some possibilities. There are many changes in corporate governance in the 1980s and 1990s, perhaps including improvements in monitoring technology. The breakdown of fixed commissions and the introduction of NASDAQ and its automated quotation 19

21 system reduce the costs of trading and so increase the liquidity of traded firms. General increases in liquidity and the efficiency of monitoring (supply conditions) are likely to expand the class of firms that are viable candidates for public equity financing. There is, however, reason to judge that changes in demand also play an important role. The broadening of the class of firms publicly traded during is not toward smaller size. If anything, the average size of newly public firms increases, and there is a general thinning of the extreme left tail of the size distribution of new lists (Table 1). Rather the broadening is toward firms with lower initial profitability and higher growth. And the key is lower profitability; these are probably firms that in the past would be judged negative net present value investments and so not viable candidates for public equity financing. What changes to give them positive value? Fama and French (2002) argue that the rise in price-earnings ratios during is largely due to declining expected stock returns, or equivalently, a lower cost of equity capital. With a lower cost of capital, less profitable firms become positive net present value projects and viable candidates for public equity financing. Finally, our sample period ( ) is relatively short, and it is reasonable to ask whether there are similar hot markets for new lists with similar growth and profitability characteristics in earlier periods. Other evidence suggests that this is not the case. Gompers and Lerner (2001) study the IPOs of the period preceding NASDAQ. Their sample is fairly complete; it is not restricted to IPOs listed on major exchanges. They find that from 1935 to 1945, there are few IPOs (typically less than ten per year). Prior to 1959, there are no hot markets; the largest number of IPOs in any year is 51. From 1959 to 1962 there is a spurt of IPOs, ranging from 122 in 1959 to 321 in There is a bigger surge from 1968 to 1972, ranging from 204 IPOs in 1971 to 683 in 1969, numbers more like those of the sustained hot market of the last 20 years of our sample. Most of the IPOs of the pre-nasdaq period are not on Compustat in their IPO year, so we do not have information about initial profitability and growth. But the firms on NASDAQ by the end of the CRSP startup period (April 1973), which we classify as seasoned, and the non-ipo new lists of the early NASDAQ years, are probably heavy with the surviving IPOs of the hot market. In the early 20